Development of a Planar Waveguide Microarray for the Monitoring

*Phone: +44 (0) 2890976535; fax: +44 (0) 2890976513; e-mail: [email protected]. ... (2) Several algal toxins are classified corresponding to ...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Development of a Planar Waveguide Microarray for the Monitoring and Early Detection of Five Harmful Algal Toxins in Water and Cultures Sara E. McNamee,† Christopher T. Elliott,† Brett Greer,† Michael Lochhead,‡ and Katrina Campbell*,† †

Institute for Global Food Security, School of Biological Sciences, Queen’s University, Stranmillis Road, Belfast BT9 5AG, United Kingdom ‡ MBio Diagnostics Inc., 5603 Arapahoe Avenue, Boulder, Colorado 80303, United States ABSTRACT: A novel multiplex microarray has been developed for the detection of five groups of harmful algal and cyanobacterial toxins found in marine, brackish, and freshwater environments including domoic acid (DA), okadaic acid (OA, and analogues), saxitoxin (STX, and analogues), cylindrospermopsin (CYN) and microcystins (MC, and analogues). The sensitivity and specificity were determined and feasibility to be used as a screening tool investigated. Results for algal/cyanobacterial cultures (n = 12) and seawater samples (n = 33) were compared to conventional analytical methods, such as high performance liquid chromatography (HPLC) and liquid chromatography tandem mass spectrometry (LC-MS/MS). Detection limits for the 15 min assay were 0.37, 0.44, 0.05, 0.08, and 0.40 ng/mL for DA, OA, STX, CYN, and MC, respectively. The correlation of data obtained from the microarray compared to conventional analysis for the 12 cultures was r2 = 0.83. Analysis of seawater samples showed that 82, 82, 70, 82, and 12% of samples were positive (>IC20) compared to 67, 55, 36, 0, and 0% for DA, OA, STX, CYN, and MC, respectively, for conventional analytical methods. The discrepancies in results can be attributed to the enhanced sensitivity and cross-reactivity profiles of the antibodies in the MBio microarray. The feasibility of the microarray as a rapid, easy to use, and highly sensitive screening tool has been illustrated for the five-plex detection of biotoxins. The research demonstrates an early warning screening assay to support national monitoring agencies by providing a faster and more accurate means of identifying and quantifying harmful toxins in water samples.

1. INTRODUCTION During recent decades, there has been an increase in the occurrences of harmful algal blooms (HAB) in many marine and freshwater systems.1 HABs are caused by the production of biotoxins belonging to three taxa groups; diatoms, dinoflagellates, and cyanobacteria. There is also an increasing number of toxic algal species and algal toxins identified with the economic losses from HABs greater than before.2 Several algal toxins are classified corresponding to their acute symptoms in humans and include amnesic shellfish poisoning (ASP) toxins, diarrheic shellfish poisoning (DSP) toxins, and paralytic shellfish poisoning (PSP) toxins. Cyanotoxins such as microcystins (MC) and cylindrospermopsin (CYN) initially found in freshwater environments are presenting in brackish and marine environments3 and have been observed to be spreading from tropical to temperate environments.4 Current methods of detection for marine biotoxins are primarily aimed for shellfish where regulations are in place within the EU (Regulation (EC) No. 853/2004) and worldwide (Codex, STAN 292-2008). No regulations are currently in place for the detection of marine biotoxins in aquatic samples. MCs are the only toxin with a guideline value of 1 μg/L in drinking water recommended by the World Health © 2014 American Chemical Society

Organization (WHO). Therefore, many of the current methods do not have the required sensitivity to act as an early warning screening method for water quality monitoring. For marine biotoxins analytical methods such as HPLC and LC-MS/MS were originally developed for their detection in shellfish as an alternative to the mouse bioassay but have been adapted for detection in algal and seawater samples for PSP, OA and DA toxins.4−12 Analytical tools such as HPLC with UV13 or MS14,15 detection have been developed for MC and LC-MS/MS for CYN detection.4,16,17 Although analytical methods are available, it is acknowledged that these methods require skilled personnel, are labor intensive, can be hindered by the lack of available analytical standards and tend to be specific to a single toxin group. Immunological assays such as ELISA18−23 and lateral flow devices24−28 have shown promise as sensitive rapid commercially available screening tools (e.g., Abraxis, Jellet Rapid Testing Ltd., Neogen). The emergence of biosensor based immunological assays have shown many Received: Revised: Accepted: Published: 13340

August 25, 2014 October 15, 2014 October 16, 2014 October 31, 2014 dx.doi.org/10.1021/es504172j | Environ. Sci. Technol. 2014, 48, 13340−13349

Environmental Science & Technology

Article

Figure 1. Multiplex MBio cartridge (A), spotting layout of slide (2 × 18 spots) (B), version 1 spotting arrangement of slide with biotoxins at three spotting concentrations (C), version 2 spotting arrangement of slide with biotoxins at optimized spotting concentrations (D) and fluorescently imaged cartridge (E). Spotting arrangement of slide including positions of controls (fluorescently labeled bovine serum albumin, buffer, GaM, and GaR antibody controls) and toxins (MC, STX, DA, OA, and CYN).

biotoxins and an Acquity UPLC I-Class system coupled to a Xevo TQ-MS mass spectrometer was used for analysis of cyanobacterial toxins (Waters, Ireland). A Waters alliance 2695 separation module HPLC system equipped with a Waters 2475 fluorescence detector (Waters, Ireland) was employed for PSP analysis. 2.2. Reagents. Phosphate buffered saline (PBS) tablets, Tween 20, bovine serum albumin (BSA) and Guillards medium were purchased from Sigma-Aldrich (Dorset, U.K.). Alexa Flour 647 goat antirabbit IgG (GaR) and Alexa Flour 647 goat antimouse IgG (GaM) antibodies were purchased from Invitrogen Ltd. (Paisley, Scotland). Saxitoxin (STX), decarbamoyl saxitoxin (dcSTX), neosaxitoxin (NEO), decarbamoyl neosaxitoxin (dcNEO), gonyautoxins (GTX1/4, GTX2/3, GTX 5), decarbamoyl gonyautoxins (dcGTX2/3), N-sulfocarbamoylgonyautoxin-2 and -3 (C 1/2), okadaic acid (OA), dinophysistoxin 1 (DTX1), dinophysistoxin 2 (DTX2) and domoic acid (DA) were purchased from the National Research Council of Canada (Halifax, Canada). Microcystin leucinearginine (MC-LR) was purchased from Enzo Life Sciences (Exeter, U.K.). Cylindrospermopsin (CYN) was purchased from n’Tox (France). BG11, Jaworski’s, L1 media and Alexandrium tamarense 1119/20 (nontoxic culture) were all purchased from the Culture Collection of Algae and Protozoa (CCAP, Oban, Scotland), Microcystis aeruginosa 91342 and 91094 were received from the Laboratory of Ecotoxicology, Genomics and Evolution (LEGE) at the Centre of Marine and Environmental Research (CIIMAR, Porto, Portugal). Anabaena spp., Microcystis aeruginosa 84/1, Planktothrix rubescens 9316, 9743, 9707, and Planktothrix argardhii 126 were kindly donated by the University of Portsmouth (UK). Cylindrospermopsis raciborskii AWT205, Alexandrium fundyense GTCA28, and Prorocentrum lima PL2V were kindly donated from Centre

advantages including high sensitivity, low detection limits, portability, specificity, and robustness. Two of the most promising biosensor methods are based on surface plasmon resonance (SPR)23,29−39 and planar waveguide14,40 but as single toxin analysis. Multi toxin analysis using SPR, chemiluminescence, flow cytometry, and microfluidic detection platforms have been described but are currently restricted for portability as are laboratory based tools.41−46 The MBio biosensor is an innovative technique based on planar waveguide with fluorescence detection showing promise for portable multiplex detection of biotoxins based on its application in the clinical setting for HIV and coinfections.47 The cartridge is composed of a plastic slide on which a microarray of toxin conjugates are spotted (Figure 1). Light directed at the bottom of the cartridge via the reader allows excitation of the fluorophores of the labeled antibody, enabling the sensitive detection of binding events. When no toxin is present, the antibody binding to the surface is optimal and a high fluorescent signal is generated. When toxin is present in the sample, the antibody binds to the toxin, thereby diminishing the fluorescent signal. Previous authors have described assays utilizing the MBio biosensor for PSP toxins and MC but in a single assay format.14,48 The aim of this research was to develop a multiplex assay for the simultaneous detection of five groups of harmful algal toxins. The method was compared with conventional analytical methods and evaluated using seawater samples collected in Europe, in addition to algal/cyanobacterial cultures.

2. MATERIALS AND METHODS 2.1. Instrumentation. The SnapEsi LS system was supplied by MBio Diagnostics Inc. (Boulder, Colorado, U.S.A.). An Acquity UPLC system coupled to a Quattro premier XE mass spectrometer was used for analysis of marine 13341

dx.doi.org/10.1021/es504172j | Environ. Sci. Technol. 2014, 48, 13340−13349

Environmental Science & Technology

Article

The cartridge was preconditioned with MBio assay buffer (150 μL) for 10 min followed by antibody: sample mix (150 μL, 1:1) for 15 min. The cartridge was washed with MBio assay buffer (2 × 150 μL, 5 min), and the secondary labeled antibodies (Alexa Flour 647 goat antirabbit IgG and Alexa Flour 647 goat antimouse IgG) were added (150 μL, 1/200 in MBio assay buffer) for 10 min. The cartridge was washed again with MBio assay buffer (2 × 150 μL, 5 min). Fifteen min assay: Antibody: sample mix (150 μL, 1:1) was applied to the cartridge for 7.5 min followed by the secondary labeled antibodies (150 μL, 1/200 in MBio assay buffer) for 7.5 min. For each assay, the cartridge was read immediately on the MBio SnapEsi reader. A series of exposures to gain dynamic range were collected and the final numerical output was scaled that it corresponded to 100 ms exposure. 2.6.4. Evaluation of Seawater Matrix Effects. Calibration curves were prepared in MBio assay buffer and extracted seawater at concentrations across the full dynamic range of each curve for a full evaluation of matrix effects. Final concentrations of calibrants were as follows; DA (0.5−20 ng/mL), OA (0.5−4 ng/mL), STX (0.02−0.5 ng/mL), MC (0.2−10 ng/mL), and CYN (0.5−5 ng/mL). Fresh, natural seawater (salinity 33−34 ppt) was obtained from Strangford Lough (Co. Down, NI). Seawater (1 L) was extracted using the method as described previously for sample preparation. 2.6.5. Limit of Detection. The limit of detection (LOD) is defined as the lowest concentration of toxin in a sample that can be detected. The theoretical LOD for this research was defined as the 20% inhibitory concentration (IC20), which is the concentration of toxin required to reduce the response by 20% compared to the response when no toxin is present (100% binding). This was calculated from an average of calibration curves analyzed during this research (n = 3). 2.7. Comparison with Conventional Analytical Methodology. Four individual analytical methods were employed as follows: 2.7.1. LC-MS/MS: OA (DTX1 and DTX2). Samples were analyzed for OA, DTX1, and DTX2 by LC-MS/MS following the standard operating procedure by the European Union reference laboratory for marine biotoxins (EU-RL-MB, Version 4, Jul 2011). 2.7.2. LC-MS/MS: DA. Samples were analyzed for DA by LCMS/MS following the standard operating procedure by the European Union reference laboratory for marine biotoxins (EU-RL-MB, Version 1, Feb 2010). 2.7.3. LC-MS/MS: MC and CYN. An Acquity UPLC HSS T3 column (100 × 2.1 mm2 id, 1.8 μm particle size) at a column temperature of 45 °C and flow rate of 0.45 mL/min was used. The mobile phases were water (0.1% formic acid) and acetonitrile. The gradient used was based on that published49 with some modifications to achieve optimal separation. The acetonitrile was held initially at 2% for 1 min, followed by an increase to 70% over 9 min, then washed for 1 min at 85% before returning to 2% for 1 min re-equilibration time. 2.7.4. HPLC: STX (and Analogues). Samples were analyzed for STX and analogues (C1/2, GTX2/3, STX, GTX1/4, dcNEO, NEO, dcGTX2/3, dcSTX, GTX5) by HPLC-FLD following a modification of the precolumn oxidation Lawrence AOAC official method.50 Modifications to this method included that the toxins were extracted from the seawater and not shellfish samples and as such the C18 cleanup stage was not required.

National de la Recherche Scientifique (CNRS, France), University of Maine (U.S.A.) and the Instituto Español de Oceanografia (Spain) respectively. 2.3. Algal Culturing. All glassware and media were sterilized prior to use at 121 °C for 30 min. Culturing of algae and cyanobacteria was performed in a UV3 HEPA PCR cabinet to maintain a sterile environment. All cultures were grown at 20 °C (±5 °C) with a 12:12 h light: dark cycle with a light intensity of 170 μmol/m2/s (Alexandrium and Prorocentrum), 116 μmol/m 2 /s (Microcystis) or 5 μmol/m 2 /s (Cylindrospermopsis, Anabaena and Planktothrix). 2.4. Seawater Sample Collection. Seawater samples were collected during 2009−2011 from a number of sites across Europe as part of the MIDTAL (Microarrays for the detection of toxic algae) EU project. The seawater samples were filtered through a 0.45 μm nitrocellulose filter until the filter clogged (approximately 0.5−2 L) depending on level of phytoplankton in the water. The filter was frozen in an eppendorf tube at −20 °C and shipped to Queen’s University, Belfast for further extraction and analysis for biotoxins. 2.5. Sample Preparation. Cell lysis was thoroughly examined in previous research.43 Algal/cyanobacterial (toxic, nontoxic and unknown toxicity) cultures (n = 12) as well as seawater samples (n = 33) were prepared as described previously using the optimized cell lysis method.43 In brief, cultures (10 mL) and seawater samples (0.5−2 L) were filtered through 0.45 filters and frozen at −20 °C until use. The filter was thawed and toxins extracted using 2 mL deionized water (with the exception that MIDTAL partner NUIG extracted in 5 mL) by vortexing for 20 s, mixing end over end for 20 min, bead beating with 0.5 mm glass beads (1 g) for 20 min on a merris minimix shaker (Merris Engineering Ltd., Ireland) followed by centrifugation at 3000g for 10 min. Finally, the supernatant was filtered using a 0.45 μm syringe filter. Samples were analyzed using the MBio biosensor and analytical methods using the same extract. 2.6. MBio Methodology. 2.6.1. Preparation of Immunological Reagents. The production of MC-LR transferrin (MCLR-TR),14 neosaxitoxin-ovalbumin (NEO-OVA),48 and cylindrospermopsin-ovalbumin23 were previously described. Domoic acid-ovalbumin29 and okadaic acid-ovalbumin30 were prepared in a similar manner to that described previously for the production of antibodies to these toxins. The preparation of immunogens and the production of the STX antibody,31 OA antibody,30 DA antibody,29 MC antibody,39 and CYN antibody23 were previously published. 2.6.2. Cartridge Spotting. Microarrays were printed using a Bio-Dot AD3200 robotic arrayer equipped with a Bio-Jet print head capable of dispensing 20 nL droplets with spot diameters approximately 0.5 mm. Toxin conjugates for DA, OA, and CYN were spotted (version 1 microarray) at three concentrations (10, 50, and 100 μg/mL) in replicates of two (Figure 1C). Following optimization one concentration of toxin was selected (version 2 microarray, Figure 1D) and spotted in replicates of four. Fluorescently labeled protein conjugates and antispecies antibodies were also spotted as controls (Figure 1D). After printing, the waveguide arrays were rinsed with a protein-based blocking agent and dried by centrifugation. 2.6.3. Assay Protocol. Assays were carried out on a specially designed cartridge rack, angled for optimum flow rate enhancing fluid flow. Flow rate was calculated as approximately 60 μL/min. Two assay protocols were designed comprising of a 45 and 15 min assays. Each protocol is described; 45 min assay: 13342

dx.doi.org/10.1021/es504172j | Environ. Sci. Technol. 2014, 48, 13340−13349

Environmental Science & Technology

Article

Table 1. Comparison of the Midpoint of the Curve (IC50) and Dynamic Range (IC20−IC80) of Each Biotoxin When Examining Single and Multi Assays, 45 and 15 min Assay Times and Seawater Matrix Effect for the MBio Biosensora biotoxin DA

OA (and analogues)

STX (and analogues)

CYN

MC

system single multi multi multi LC/MS-MS single multi multi multi LC/MS-MS single multi multi multi HPLC single multi multi multi LC/MS-MS single multi multi multi LC/MS-MS

matrix MBio assay MBio assay MBio assay Seawater Seawater MBio assay MBio assay MBio assay seawater seawater MBio assay MBio assay MBio assay seawater seawater MBio assay MBio assay MBio assay seawater seawater MBio assay MBio assay MBio assay seawater seawater

assay time (min)

mid-point IC50 (ng/mL)

dynamic range IC20−IC80 (ng/mL)

45 45 15 15 3 45 45 15 15 8.5 45 45 15 15 14 45 45 15 15 12 45 45 15 15 12

0.13 1.93 1.83 1.37 N/A 0.24 0.42 1.06 0.96 N/A 0.04 0.06 0.06 0.09 N/A 0.23 0.19 0.39 0.26 N/A 1.48 1.25 1.46 1.31 N/A

0.03−1.07 0.56−5.72 0.63−5.04 0.37−4.43 2.5b 0.13−0.66 0.22−0.78 0.52−2.00 0.44−1.94 2.5b 0.02−0.06 0.02−0.15 0.02−0.18 0.05−0.15 2.5b 0.07−0.56 0.08−0.52 0.20−0.76 0.08−0.72 10c 0.50−3.73 0.40−3.41 0.48−3.78 0.40−3.35 10c

buffer buffer buffer

buffer buffer buffer

buffer buffer buffer

buffer buffer buffer

buffer buffer buffer

a Analytical methods used included both LC-MS/MS and HPLC. bDenotes the sensitivity of these methods are based on the lowest quantifiable peak observed. cDenotes the lowest standard analyzed in the experiment (no peaks in samples observed less than this).

3. RESULTS AND DISCUSSION The present study outlines for the first time the use of this technology to detect biotoxins in a multiplex format. The increasing concerns due to biotoxins in relation to both food and water quality safety has necessitated the need for rapid, sensitive, portable, high throughput, and multiplex detection systems. 3.1. Assay Protocol. Following the work of Lochhead et al. (2011)47 and Meneely et al. (2013)48 the initial optimization for the MBio assay was performed as a single 45 min assay for each toxin group. This allowed the determination of optimal spotting concentration of toxin conjugates, antibody dilutions and to evaluate interference between antibodies and toxins immobilized onto the cartridge. Antibodies were optimized to give a response 300−500 fluorescence units and inhibition was assessed. The spotting and optimization of the microarray for STX and MC were discussed in full in Meneely et al. (2013)48 and Devlin et al. (2013) to give optimized concentrations of 50 μg/mL using NEO-OVA and 100 μg/mL using MC-LR-TR. For DA, OA, and CYN, three spotting concentrations were assessed at 10, 50, and 100 μg/mL on version 1 of the microarray (Figure 1C). Inhibition and sensitivity were optimum for DA and OA at a spotting concentration of 10 μg/mL while 50 μg/mL was necessary for CYN. This was because the CYN antibody had to be diluted to 1/6000 to show inhibition, as such the response was below the 300−500 fluorescence units required at 10 μg/mL. The final optimum spotting layout of all five toxins can be seen in Figure 1(D). Each antibody was optimized at dilutions of 1/6000 for DA, OA, STX, and CYN and 1/40 000 for MC giving a response between 300 and 500 fluorescence units as five single assays. It

was ascertained that nonspecific binding was not occurring between the binding proteins. All five toxins were then incorporated into a multi assay at 45 min. Finally, the assay was reduced to a 15 min multi assay to be more applicable for use in the field. Antibody dilutions were reoptimized (1/2000 for OA and CYN, 1/5000 for DA, 1/8000 for STX and 1/10 000 for MC) in addition to standards for performance and sensitivity. The midpoint and dynamic range were determined for each assay using BIAevaluation software (Table 1). Cartridges can be batch processed, with up to 32 run in parallel showing stability up to 60 min with average coefficient of variation (CVs) no greater than 1.5% on reading immediately and 60 min later. Antibody specificity was assessed in previous publications for the MC, CYN, OA, and STX antibodies.14,23,33,48 Crossreactivity data was determined by calculating the midpoint concentration (IC50) of each toxin curve as a percentage relative to the main standard. For PSP toxins these were STX 100%, dcSTX 49%, GTX5 33%, dcGTX2/3 28%, C1/2 25%, dcNEO 8%, GTX2/3 7%, NEO 4%, and GTX1/4 1250 nd 35.0 261.2 16.8 189.9

total

analogue

nd 730.0

nd C1/2: 198.0; GTX2/3: 158.8; STX: 47.5; GTX1/4: 284.5; NEO: 39.2; GTX5:2.0 OA: 112.5; DTX1:27.2 nd MC-LR: 228.0; MC-LA: 20 MC-LR: 4.9 CYN: 1283.2 nd MC-YR: 2.6 MC-LR: 1.4; MC-YR: 10 nd MC-YR: 8.3

139.7 nd 248.0 4.9 1283.2 nd 2.6 11.4 nd 8.3

Samples were analysed for all toxins but results show only toxins detected. nd denotes not detected.

(47.5, 39.2, and 2.0 ng/mL respectively). Prorocentrum lima PL2V is a known toxin producer of OA and DTX1,8,51,52 which was confirmed by both the MBio biosensor (108.0 ng/mL) and LC-MS/MS (139.7 ng/mL) with the LC-MS/MS detecting OA (112.5 ng/mL) and DTX1 (27.2 ng/mL). Both the MBio biosensor (>1250 ng/mL) and LC-MS/MS (1283.2 ng/mL) detected Cylindrospermopsis raciborskii AWT205 as positive for CYN production. The Planktothrix cultures and Microcystis 84/ 1 are producing MC as measured by the MBio biosensor. These results were confirmed by LC-MS/MS with the only discrepancy being for Planktothrix rubescens 9707, positive for MC by the MBio biosensor but negative by LC-MS/MS. This may be because of the cross-reactivity profile of the MC antibody, which may be detecting a MC variant that is not included in the LC-MS/MS profile. The Anabaena spp. was negative for all toxins on the MBio assay and LC-MS/MS. 3.4.2. Seawater samples. Seawater samples (n = 33) were analyzed by the MBio biosensor and analytical methods (Table 3). DA: There were five samples (MBA 5, SZN 5C, NUIG 39, NUIG 42, and INT 28) showing DA detection by the MBio biosensor (result greater than IC20) with no DA present by LCMS/MS. For samples MBA 5 and SZN 5C this was due to sensitivity with the MBio biosensor showing much lower sensitivity. Sample INT 28 showed a possible peak for DA by LC-MS/MS but too much background noise meant the sample could not be accurately quantified. For DA only one toxin analogue is available as an analytical standard but there are various isoforms that occur of DA which would potentially be detected by the antibody. The biosensor assay will cumulatively detect all toxins whereby the LC-MS/MS will detect and identify only DA. At this level of detection, the MBio is more sensitive and whereas more than one analogue could be present below the detection level of the LC-MS/MS, giving a cumulative response greater than the detection level by MBio. This explains why no DA was detected in samples NUIG 39 and 42 by LC-MS/MS when the MBio showed approximately 4 ng/filter. Where DA was detected by LC-MS/ MS, the MBio results were positive (+), but in general it appears that the MBio may offer enhanced sensitivity for detection.

dynamic range for each toxin in the 15 min assay protocol (Table 1). Nonetheless, the analysis of seawater samples was carried out with a matrix matched seawater calibration curve to ensure accurate determination of toxin content and to maintain minimal matrix effects. 3.3. Limit of Detection. The theoretical LOD for these assays was defined as the IC20. The LOD (Table 1) for all toxins have been established for each system, assay time, and matrix. The LOD when analyzing single toxin as a 45 min assay (MBio assay buffer) were 0.03, 0.13, 0.02, 0.07, and 0.50 ng/ mL for DA, OA, STX, CYN, and MC, respectively. When comparing these LODs but in a multi system (45 min), the LOD for both STX and CYN remained similar, doubled for OA and decreased slightly for MC. DA was the only toxin where a substantial change in the IC20 was observed increasing to 0.56 ng/mL compared with 0.03 ng/mL (single assay). When the assay time was shortened to 15 min, the LOD remained relatively similar for DA, STX, and MC while OA and CYN doubled. Finally, when using matrix extracted seawater in the LOD remained similar for MC (0.40 ng/mL), decreased for DA (0.37 ng/mL), CYN (0.08 ng/mL), and OA (0.44 ng/mL) but increased for STX (0.05 ng/mL). MC is the only toxin with a regulatory limit (1 μg/L = 1000 ng/mL) in place for water. The LOD for MC using the MBio biosensor is approximately half this at 0.4 ng/mL therefore sensitivity of all toxins are low enough allowing use as an early warning detection tool. 3.4. Sample Analysis and Comparison with Analytical Methods. 3.4.1. Algae and Cyanobacteria Sample Analysis. Algal/cyanobacterial cultures (n = 12; marine = 3 and freshwater = 9) were analyzed using the MBio biosensor (Table 2). The known toxic cultures (GTCA28, PL2V, 91094, AWT205) were all shown to be producing toxins while the nontoxic culture (1119/20) was negative. The toxic Alexandrium f undyense culture GTCA2848 showed 70.1 ng/mL of PSP toxins by the MBio biosensor compared to 730.0 ng/mL (457 STXeqs/ml) by HPLC-FLD. Differences detected were due to the cross-reactivity profile of the antibody, but the sensitivity offered by the microarray allows for lower levels of STX to be determined compared to the HPLC method. The HPLC data confirmed that this culture was producing mostly GTX1/4 (284.5 ng/mL), C1/2 (198.0 ng/mL), and GTX2/3 (158.8 ng/ mL) as well as lower concentrations of STX, NEO, and GTX5 13344

dx.doi.org/10.1021/es504172j | Environ. Sci. Technol. 2014, 48, 13340−13349

20 11.7 >20 nd

5C 19C 22B 10C 14C

Gulf of Naples, Italy

4.7

42 1.0 7.0 nd >20 20 >20

>50 50

>20 >20 >20 >20

16B 18C 20B 23C

2A

>20

10B

3.1

5

Arcachon Bay, France

Marine Biological Association (MBA), Plymouth, U. K. Stazione Zoologica “A. Dohrn” di Napoli (SZN), Naples, Italy

(ng/filter)

sampling site

partner

sample no.

MBio

+ + − + − + +

+

+ − +

+

+ + + +

2.4 6.0 nd 54.4 nd 287.5 184.9

nd

53.3 nd nd

13.0

1845.0 1182.3 23.1 7.0

208.5

nd 7.7

− +

+

nd 29.8 6.8 104.0 nd

nd

LC-MS/ MS (ng/filter)

+ + + + −

+

+/−

DA

4.2 nd nd 2.0 nd 1.6 2.7

2.8

4.3 nd 4.8

>15

>6 >6 >6 >6

>6

nd >6

1.7 1.9 1.9 0.9 2.2

2.0

(ng/filter)

MBio

+ − − + − + +

+

+ − +

+

+ + + +

+

− +

+ + + + +

+

+/−

97.35 289.08 84.36 1460.07

OA: 6.87 nd nd OA: 3.63 nd nd nd

nd

OA: 25.3 DTX2:14.2 nd nd nd

OA: OA: OA: OA:

OA: 134.9 DTX2:42.0

nd OA: 283

nd nd nd OA: 3.52 OA: 4.62

nd

LC-MS/MS (ng/filter)

OA (and analogues)

0.1 0.2 nd 0.2 0.1 0.3 >0.4

>1

0.1 1

0.4

0.4

nd >0.4

0.2 >0.4 0.1 0.1 >0.4

6

>6 6

>6 1.9 >6 >6

(ng/filter)

+

+ − +

+ + + +

+/− 11.64 4.12 67.35 151.31

33 18 15

OA: 147.54

OA: 140.73 nd OA: 43.41

OA: OA: OA: OA:

LC-MS/MS (ng/filter)

OA (and analogues) MBio

33 23 10

>0.4

nd nd >0.4

>0.4 nd